2016 Postmortem
Related: About this forumWhy is Nate Silver giving Rasmussen 4x and ARG 2x the weight of other pollsters?
Right now, on his front page table, every other pollster gets one set of numbers in his average. Whereas two know right-biased pollsters, Rassmussen and ARG have multiple polls included.
He's mixed a few state-wide polls in with nationwide polls.
The 10/12 High Point data is for N. Carolina through 10/10, not nationwide and not yesterday.
The data for ARG is from 2 states, not nationwide.
And Google Consumer Surveys shows a single post-debate survey on their site dated 10/4, before the lies were exposed and the good economic news came. Their consumer survey model gives responders free access to something in exchange for the survey. If that is their model for their election survey, which I don't see described on their site, then the nature of the free information access clearly influences the type of person who responds.
When you pull out the state-wide polls and the (suspect in my mind) Google Survey, and look at the rest, the national post-debate average *net change* is +2.7 and the median +1.85 swing to Romney.
When you pull out the state-wide polls and the Google Survey, and look at the rest, Obama's current standing nationwide averages +1.1.
oswaldactedalone
(3,491 posts)"Because they're reliable and accurate pollsters?"
DemocratSinceBirth
(99,710 posts)http://fivethirtyeight.blogs.nytimes.com/2010/11/04/rasmussen-polls-were-biased-and-inaccurate-quinnipiac-surveyusa-performed-strongly/
progressivebydesign
(19,458 posts)Them??? Did you not see the tweet from PPP about the mass email that (R)Assmussen sent out yesterday???
It said, basically "The DVD 2016 can change the course of the election. Get your Copy today!!"
MrMickeysMom
(20,453 posts)According to an expert Detroit-guy- all around smart guy in my family, Nate knows what the fuck he's talking about, so when he balances stuff, we have a tendency to agree with what Nate balances.
BTW, Nate developed all his math models from all of his study of baseball (because of ALL of the numbers in baseball)
Beside, Jennifer Granholm likes him, so.... harumph!
And, Go Tigers...
fredamae
(4,458 posts)in polls? Think about Who Owns these Polling Corps!
Poll$ are Only as reliable as the per$on who Paid$ for them is hone$t.
Remember that and don't allow for too much reliance upon their credibility.
A Lot of folks place too much faith in their credibility. Remember there is Much to be gained by "skewing" the results to meet a desired outcome---to snag many who simply follow without verification.
Just make your own decisions based upon the reality that is happening all around you thoroughly Vet every candidate right down to your City Council!
oswaldactedalone
(3,491 posts)(quoting from my favorite movie, Fletch) when the polls have good news for Obama everyone loves the poll and the pollster. When the news turns bad, the polls are meaningless and the pollster dishonest. Can't have it both ways people.
ProSense
(116,464 posts)"...when the polls have good news for Obama everyone loves the poll and the pollster. When the news turns bad, the polls are meaningless and the pollster dishonest. Can't have it both ways people."
...anyone take ARG seriously? No one posted its poll numbers prior to the debate. Rasmussen is always posted with caveats acknowledging the right-leaning bias.
DemocratSinceBirth
(99,710 posts)Obviously the president was doing better on October 2nd than he is doing now but I don't think his and our prospects are as dire now as some of the prognosticators said his prospects were rosy then. It's not as if he had a large cushion on October 2nd. If it was a fifteen round boxing match the president was up 7-6 in the thirteenth round and now it's all tied up going into the fifteenth with the president holding a small advantage.
boxman15
(1,033 posts)Its results lean to the right as we all know, but it still picks up trends over time. I have no problem with Nate using it.
DemocratSinceBirth
(99,710 posts)I have no doubt Rasmussen is legitimate. He was one of the founders of ESPN. Sometimes I question whether small outfits like ARG even make any calls. He's not a robocaller and claims to call cell phones so that gets expensive.
TDale313
(7,820 posts)538 has been pretty accurate, even when I don't like the numbers.
ProSense
(116,464 posts)acknowledges the limitations of his model, which is that he has to include polls even when he questions the data.
http://fivethirtyeight.blogs.nytimes.com/2012/10/12/oct-12-romney-debate-gains-show-staying-power/
magical thyme
(14,881 posts)but then doubles its impact.
He admits Rasmussen leans right and then quadruples its impact.
It doesn't make sense to me. I don't care how accurate he has been in the past. I want to understand why he is weighting admittedly skewed results more highly than nonskewed results. And I want to understand why he is mixing state results with national results. One strikes me as mixing rotten apples into the bag. The other mixing apples and oranges.
I passed my pre-med statistics course with close to 100%, so I have an idea of how surveys and statistics work and how they can be skewed. I want to understand why he is doing something that appears to skew things in one particular direction. There may well be a good rationale for it. I'm trying to understand what it is.